Master Thesis

PUBDB-2025-05593

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Search for long-lived supersymmetric decays in CMS using machine learning methods

Vashishtha, S. (Corresponding author)DESY* ; Jolie, J. (Thesis advisor) ; Melzer-Pellmann, I. (Thesis advisor)DESY* ; Desch, K. (Thesis advisor)Extern*

2025

Abstract: As the search for physics beyond the Standard Model continues, this thesis presentsan analysis in the long-lived particle (LLP) supersymmetry sector. A Boosted DecisionTree (BDT) classifier was developed to enhance the search for long-lived supersymmetricpartner particles of the tau lepton (stau) particles in the muon-hadronic tau channel at theCMS detector in LHC, motivated by Gauge Mediated Supersymmetry Breaking (GMSB)scenarios. The signal region is characterized by the staus decaying to a muon and hadronictau. These displaced topologies were analyzed with the help of machine learning tools. Usingsimulated Run 2 CMS data, a BDT was constructed, including input feature selection, eventweighting, cross-validation, and model optimization. It demonstrates strong performance,achieving up to 90% signal efficiency while maintaining background misidentification ratesbelow 10−4, depending on the chosen working point. These results demonstrate the BDT’spotential for deployment in ongoing and future searches for long-lived particles at the LHC.Future work will address systematic uncertainties and integrate the BDT into the full eventselection workflow to further improve sensitivity to GMSB-inspired new physics.

Note: Masterarbeit, University of Cologne, 2025

---

Contributing Institute(s):

1. LHC/CMS Experiment (CMS)

Research Program(s):

1. 611 - Fundamental Particles and Forces (POF4-611) (POF4-611)

Experiment(s):

1. LHC: CMS

---